On Microsoft Translator’s Performance in English-Persian speech-to-text Translation: Recognizing Translation Errors and Identifying the Source of Errors

Over the past decades, the language industry has benefited from computer-aided tools; although technology has not affected interpreting to the same extent as translation, some improvements have been made in that profession as well. The desire to avoid cognitive saturation has increased interest in computer-assisted tools and speech translation systems among interpreters. However, to better understand the function of these systems, it is necessary to study their performance, identify possible errors, and evaluate output data quality. Despite worldwide interest in detecting the role of technology in interpreting, the area of computer-assisted interpreting has been quite under-researched in Iran. Therefore, in an attempt to fill this gap, this study sought to evaluate Microsoft Translator speech translation performance. Since Microsoft Translator is a free and easily accessible translation tool. The researchers aimed to identify translation errors as well as recognizing the possible source of errors. To that end, corpora of political hearings addressed at United Nation sessions, their speech-to-text translations by Microsoft Translator, and a reference translation were created and analyzed. To find answers to the first research question, Microsoft Translator errors were detected and categorized based on the component responsible for generating the errors. The second question concerned the probable causes of errors. The findings showed that Internet access, time delay, manual function of the microphone, and speaking features could lead to translation errors. The findings of this research can be a starting point for future research in the field of computer-assisted interpreting.